Piecewise temperature dependent electrical equivalent modeling of PEM fuel cell for power conditioning unit design using fuzzy clustering and hybrid optimization

Abstract

ABSTRACT Effective utilization of Polymer Electrolyte Membrane Fuel Cell (PEMFC) for power electronic applications demands a reliable model with accurate parameter estimation. Existing approaches for modeling the PEMFC using terminal characteristics are based on the assumption of uniform temperature throughout the fuel cell operation. However, stack temperature varies widely for different operating conditions. In this regard, the present work proposes a temperature dependent piecewise modeling approach by dividing the operating range into a set of piecewise regions with different model parameter characterizing each region. To achieve the same, the piecewise regions have been distributed using a fuzzy clustering approach by maintaining a compromise between precise representation of experimental data and reduced model complexity. The parameter estimation for each region is carried out using a novel hybrid optimization approach. The model can quantify the dependence of model parameters on temperature variation. With the temperature-dependent piecewise representation, the proposed model significantly outperforms the conventional uniform temperature model in representing the experimental data by achieving a fitting accuracy of 97.9% at 311 K and 99.8% at 319 K, respectively.